The following figure illustrates that everybody has it's own tagging engine.
Each server has it's own tags for it's projects.
- I (the repository hoster) control everything myself
- e.g. freshmeat lists lots of projects that are hosted somewhere else, so do they allow to tag those? -> Projects may contain different sets of tags, depending on which site you look for tags
This figure illustrates how one single tagging server could work and be connected to others.
All tags are stored on one single server (well, physically this can also be a distributed system, but at some level this should be transparent)
- tags added on e.g. sourceforge could be seen immediately on freshmeat too :)
- consistent tags for each project, no matter on which site you look at it
- even very small projects that don't use one of the participating software repositories might join the central server and thus could be tagged and found by searches
- dependence on s.b. else .. it should be well chosen who is responsible for the central tagging server (best would be some alliance between major repo-sites)
- requires unique id's for each project that participates. (could be the url like sourceforge.net/projects/osstag)
- might slow down repository-sites, as each page-view on one of those could require a connection to the server (maybe read-only replication could help here a lot)
- some repo sites might fear that this server could take away traffic from themself (could be corrected if the central-server does not offer searching via some html-interface at all, but only via participating repo-sites)